Search NASA⌕ Search

SEARCH · Search NASA

Results for “statistical inference”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 19 records
All sourcesNASA NTRSDOE OSTI
NASA NTRS2004

Multivariate Statistical Inference of Lightning Occurrence, and Using Lightning Observations

Two classes of multivariate statistical inference using TRMM Lightning Imaging Sensor, Precipitation Radar, and Microwave Imager observation are studied, using nonlinear classification neural networks as inferential tools. The very large and globally representative data sample provided by TRMM allows both training and validation (without overfitting) of neural networks with many degrees of freedom. In the first study, the flashing / or flashing condition of storm complexes is diagnosed using radar, passive microwave and/or environmental observations as neural network inputs. The diagnostic skill of these simple lightning/no-lightning classifiers can be quite high, over land (above 80% Probability of Detection; below 20% False Alarm Rate). In the second, passive microwave and lightning observations are used to diagnose radar reflectivity vertical structure. A priori diagnosis of hydrometeor vertical structure is highly important for improved rainfall retrieval from either orbital radars (e.g., the future Global Precipitation Mission "mothership") or radiometers (e.g., operational SSM/I and future Global Precipitation Mission passive microwave constellation platforms), we explore the incremental benefit to such diagnosis provided by lightning observations.

Boccippio, Dennis↗
NASA NTRS1994

Redshift data and statistical inference

Frequency histograms and the 'power spectrum analysis' (PSA) method, the latter developed by Yu & Peebles (1969), have been widely employed as techniques for establishing the existence of periodicities. We provide a formal analysis of these two classes of methods, including controlled numerical experiments, to better understand their proper use and application. In particular, we note that typical published applications of frequency histograms commonly employ far greater numbers of class intervals or bins than is advisable by statistical theory sometimes giving rise to the appearance of spurious patterns. The PSA method generates a sequence of random numbers from observational data which, it is claimed, is exponentially distributed with unit mean and variance, essentially independent of the distribution of the original data. We show that the derived random processes is nonstationary and produces a small but systematic bias in the usual estimate of the mean and variance. Although the derived variable may be reasonably described by an exponential distribution, the tail of the distribution is far removed from that of an exponential, thereby rendering statistical inference and confidence testing based on the tail of the distribution completely unreliable. Finally, we examine a number of astronomical examples wherein these methods have been used giving rise to widespread acceptance of statistically unconfirmed conclusions.

Newman, William I.↗
NASA NTRS1986

Making statistical inferences about software reliability

Failure times of software undergoing random debugging can be modeled as order statistics of independent but nonidentically distributed exponential random variables. Using this model inferences can be made about current reliability and, if debugging continues, future reliability. This model also shows the difficulty inherent in statistical verification of very highly reliable software such as that used by digital avionics in commercial aircraft.

Miller, Douglas R.↗
NASA NTRS1988

Making statistical inferences about software reliability

Failure times of software undergoing random debugging can be modelled as order statistics of independent but nonidentically distributed exponential random variables. Using this model inferences can be made about current reliability and, if debugging continues, future reliability. This model also shows the difficulty inherent in statistical verification of very highly reliable software such as that used by digital avionics in commercial aircraft.

Miller, Douglas R.↗
NASA NTRS2002

Classification image analysis: estimation and statistical inference for two-alternative forced-choice experiments

We consider estimation and statistical hypothesis testing on classification images obtained from the two-alternative forced-choice experimental paradigm. We begin with a probabilistic model of task performance for simple forced-choice detection and discrimination tasks. Particular attention is paid to general linear filter models because these models lead to a direct interpretation of the classification image as an estimate of the filter weights. We then describe an estimation procedure for obtaining classification images from observer data. A number of statistical tests are presented for testing various hypotheses from classification images based on some more compact set of features derived from them. As an example of how the methods we describe can be used, we present a case study investigating detection of a Gaussian bump profile.

Non-NASA Center↗
NASA NTRS1994

A statistical inference approach for the retrieval of the atmospheric ozone profile from simulated satellite measurements of solar backscattered ultraviolet radiation

NASA's Mission to Planet Earth (MTPE) will address important interdisciplinary and environmental issues such as global warming, ozone depletion, deforestation, acid rain, and the like with its long term satellite observations of the Earth and with its comprehensive Data and Information System. Extensive sets of satellite observations supporting MTPE will be provided by the Earth Observing System (EOS), while more specific process related observations will be provided by smaller Earth Probes. MTPE will use data from ground and airborne scientific investigations to supplement and validate the global observations obtained from satellite imagery, while the EOS satellites will support interdisciplinary research and model development. This is important for understanding the processes that control the global environment and for improving the prediction of events. In this paper we illustrate the potential for powerful artificial intelligence (AI) techniques when used in the analysis of the formidable problems that exist in the NASA Earth Science programs and of those to be encountered in the future MTPE and EOS programs. These techniques, based on the logical and probabilistic reasoning aspects of plausible inference, strongly emphasize the synergetic relation between data and information. As such, they are ideally suited for the analysis of the massive data streams to be provided by both MTPE and EOS. To demonstrate this, we address both the satellite imagery and model enhancement issues for the problem of ozone profile retrieval through a method based on plausible scientific inferencing. Since in the retrieval problem, the atmospheric ozone profile that is consistent with a given set of measured radiances may not be unique, an optimum statistical method is used to estimate a 'best' profile solution from the radiances and from additional a priori information.

Bonavito, N. L.↗
NASA NTRS2009

Statistical inference of static analysis rules

Various apparatus and methods are disclosed for identifying errors in program code. Respective numbers of observances of at least one correctness rule by different code instances that relate to the at least one correctness rule are counted in the program code. Each code instance has an associated counted number of observances of the correctness rule by the code instance. Also counted are respective numbers of violations of the correctness rule by different code instances that relate to the correctness rule. Each code instance has an associated counted number of violations of the correctness rule by the code instance. A respective likelihood of the validity is determined for each code instance as a function of the counted number of observances and counted number of violations. The likelihood of validity indicates a relative likelihood that a related code instance is required to observe the correctness rule. The violations may be output in order of the likelihood of validity of a violated correctness rule.

Engler, Dawson Richards↗
NASA NTRS1981

Cleanroom software development

The 'cleanroom' software development process is a technical and organizational approach to developing software with certifiable reliability. Key ideas behind the process are well structured software specifications, randomized testing methods and the introduction of statistical controls; but the main point is to deny entry for defects during the development of software. This latter point suggests the use of the term 'cleanroom' in analogy to the defect prevention controls used in the manufacturing of high technology hardware. In the 'cleanroom', the entire software development process is embedded within a formal statistical design, in contrast to executing selected tests and appealing to the randomness of operational settings for drawing statistical inferences. Instead, random testing is introduced as a part of the statistical design itself so that when development and testing are completed, statistical inferences are made about the operation of the system.

Dyer, M.↗
NASA NTRS2006

Exploring the Connection Between Sampling Problems in Bayesian Inference and Statistical Mechanics

The Bayesian and statistical mechanical communities often share the same objective in their work - estimating and integrating probability distribution functions (pdfs) describing stochastic systems, models or processes. Frequently, these pdfs are complex functions of random variables exhibiting multiple, well separated local minima. Conventional strategies for sampling such pdfs are inefficient, sometimes leading to an apparent non-ergodic behavior. Several recently developed techniques for handling this problem have been successfully applied in statistical mechanics. In the multicanonical and Wang-Landau Monte Carlo (MC) methods, the correct pdfs are recovered from uniform sampling of the parameter space by iteratively establishing proper weighting factors connecting these distributions. Trivial generalizations allow for sampling from any chosen pdf. The closely related transition matrix method relies on estimating transition probabilities between different states. All these methods proved to generate estimates of pdfs with high statistical accuracy. In another MC technique, parallel tempering, several random walks, each corresponding to a different value of a parameter (e.g. "temperature"), are generated and occasionally exchanged using the Metropolis criterion. This method can be considered as a statistically correct version of simulated annealing. An alternative approach is to represent the set of independent variables as a Hamiltonian system. Considerab!e progress has been made in understanding how to ensure that the system obeys the equipartition theorem or, equivalently, that coupling between the variables is correctly described. Then a host of techniques developed for dynamical systems can be used. Among them, probably the most powerful is the Adaptive Biasing Force method, in which thermodynamic integration and biased sampling are combined to yield very efficient estimates of pdfs. The third class of methods deals with transitions between states described by rate constants. These problems are isomorphic with chemical kinetics problems. Recently, several efficient techniques for this purpose have been developed based on the approach originally proposed by Gillespie. Although the utility of the techniques mentioned above for Bayesian problems has not been determined, further research along these lines is warranted

Pohorille, Andrew↗
NASA NTRS2011

Space-Time Data fusion for Remote Sensing Applications

NASA has been collecting massive amounts of remote sensing data about Earth's systems for more than a decade. Missions are selected to be complementary in quantities measured, retrieval techniques, and sampling characteristics, so these datasets are highly synergistic. To fully exploit this, a rigorous methodology for combining data with heterogeneous sampling characteristics is required. For scientific purposes, the methodology must also provide quantitative measures of uncertainty that propagate input-data uncertainty appropriately. We view this as a statistical inference problem. The true but notdirectly- observed quantities form a vector-valued field continuous in space and time. Our goal is to infer those true values or some function of them, and provide to uncertainty quantification for those inferences. We use a spatiotemporal statistical model that relates the unobserved quantities of interest at point-level to the spatially aggregated, observed data. We describe and illustrate our method using CO2 data from two NASA data sets.

carbon dioxide↗
NASA NTRS2006

Using the Bootstrap Method for a Statistical Significance Test of Differences between Summary Histograms

A new method is proposed to compare statistical differences between summary histograms, which are the histograms summed over a large ensemble of individual histograms. It consists of choosing a distance statistic for measuring the difference between summary histograms and using a bootstrap procedure to calculate the statistical significance level. Bootstrapping is an approach to statistical inference that makes few assumptions about the underlying probability distribution that describes the data. Three distance statistics are compared in this study. They are the Euclidean distance, the Jeffries-Matusita distance and the Kuiper distance. The data used in testing the bootstrap method are satellite measurements of cloud systems called cloud objects. Each cloud object is defined as a contiguous region/patch composed of individual footprints or fields of view. A histogram of measured values over footprints is generated for each parameter of each cloud object and then summary histograms are accumulated over all individual histograms in a given cloud-object size category. The results of statistical hypothesis tests using all three distances as test statistics are generally similar, indicating the validity of the proposed method. The Euclidean distance is determined to be most suitable after comparing the statistical tests of several parameters with distinct probability distributions among three cloud-object size categories. Impacts on the statistical significance levels resulting from differences in the total lengths of satellite footprint data between two size categories are also discussed.

Xu, Kuan-Man↗
NASA NTRS1983

Quantile Data Analysis of Image Data

Quantile data analysis and functional statistical inference methods are introduced and applied to provide representations of spectral data which may lead to simple statistical discriminators effective for the estimation of ground truth from satellite spectral measurements. To estimate the ground truth of a pixel, the probability of each possible ground truth is estimated, given observed (estimated) quantile theoretic statistical characteristics of the multispectral image data corresponding to the pixel and its neighboring pixels. A strategy for determining which statistical characteristics discriminate best is described. Results are reported of quantile data analysis of an extensive collection of training files of image data.

Parzen, E.↗
NASA NTRS1964

Stability in signal detection problems.

Stability in signal detection and measurement based on statistical inference made from values of functional on sample space

STATISTICAL COMMUNICATION THEORY↗
NASA NTRS1992

An early estimate for the size of cycle 23

The maximum amplitude for the next sunspot cycle is estimated on the basis of inferred statistical trends and associations from the last 12 cycles. Specific attention is given to the extrapolation of the inferred long-term upward trend against time of the maximum amplitude RM which is the maximum value of the smoothed sunspot number for a given cycle. Also examined in depth are the apparent inherent differences found in even- and odd-numbered sunspot cycles. RM is plotted vs cycle number, and RM is plotted against RM for odd-following cycle pairs. RM values are shown to be greater than 110.6 for six of the last six sunspot cycles, and the trend suggests that cycle 23 should have RM of more than 150 particularly in light of the fact that the odd-following cycle is always larger than the even cycle. Evidence is shown to predict an RM of at least 176.4, making cycle 23 one of the largest of the modern era.

Wilson, Robert M.↗
NASA NTRS2017

Modular Spectral Inference Framework Applied to Young Stars and Brown Dwarfs

In practice, synthetic spectral models are imperfect, causing inaccurate estimates of stellar parameters. Using forward modeling and statistical inference, we derive accurate stellar parameters for a given observed spectrum by emulating a grid of precomputed spectra to track uncertainties. Spectral inference as applied to brown dwarfs re: Synthetic spectral models (Marley et al 1996 and 2014) via the newest grid spans a massive multi-dimensional grid applied to IGRINS spectra, improving atmospheric models for JWST. When applied to young stars(10Myr) with large starpots, they can be measured spectroscopically, especially in the near-IR with IGRINS.

K↗
NASA NTRS2011

Use of Minute-by-Minute Cardiovascular Measurements During Tilt Tests to Strengthen Inference on the Effect of Long-Duration Space Flight on Orthostatic Hypotension

Typical methodology for evaluating the effects of spaceflight on orthostatic hypotension (OH) has been survival analysis of tolerance times from 80 head-up tilt tests. However when scheduled test durations are short, there may not be enough failures to allow survival analysis to adequately estimate and compare the effects of flight phase (e.g. pre-flight, number of days post-flight), flight duration, and their interaction, as well as interactions with effects of interventions or countermeasures. The problem is exacerbated in the presence of a repeated measures design, in which subjects participate in tilt tests during various flight phases. Here we show how it is possible to dramatically improve the efficiency of statistical inference in this setting by making use of the additional information contained in minute-by-minute observations of cardiovascular parameters thought to be reflective of progression towards presyncope during tilt testing. Methods: We retrospectively examined operational tilt test (OTT; 10 -min 80 head-up tilt) data from 20 International Space Station (ISS) and 66 Shuttle astronauts 10 d before launch (L-10), on landing day (R+0) and during recovery (R+1, R+3, R+6-10) depending on the level of participation. Data from 5 ISS astronauts tested on R+0 or R+1 who used non-standard countermeasures were excluded. In addition to OTT survival time, 8 cardiovascular parameters (CP: heart rate, systolic, diastolic, and mean arterial blood pressure, pulse pressure, stroke volume, cardiac output, and total peripheral resistance) that might be predictive of progression towards presyncope were measured every minute of each OTT. Statistical analysis was predicated on a two ]stage model of causation. In the first stage, flight duration and time from landing affect the astronauts' degree of OH, which is manifested in the time trends and variation of the above CPs during OTTs. In the second stage, the behavior of these parameters directly affects OTT survival time. Actual analysis proceeded in the opposite direction. First we identified those CPs or linear combinations that best predicted OTT survival regardless of what spaceflight conditions led to OTT completion or presyncope. From these, we calculated a summary statistic (one per OTT) that best predicted survival. We then used mixed ]model regression analysis to relate changes in the summary statistic to flight phase and duration. Inference on the effects of phase, duration, and their interaction on OH follows directly from this second analysis. Results: A linear combination (W) of diastolic blood pressure (DBP) and stroke volume (SV) was found to be the best predictor of OTT survival using the complete data set of minute-by-minute observations of CPs for each OTT. Furthermore, the log-transformed standard deviation of W (Z = log SW) was found to be a strong predictor of survival in the reduced data set consisting of one observation per OTT. In other words, this measure of variability of W during an OTT was the best indicator of whether or not the subject could complete the 10-min test, with higher variability (i.e. higher values of Z) being associated with greater probability of failure. In the mixed-model regression analysis where Z was now treated as a outcome with flight phase and duration groups (ISS and STS) as predictors, we found that there was a significantly more variability in W (higher values of Z) for both groups at R+0, but with no evidence of an interaction until R+3, when the ISS group still had inflated variability, but not the STS group. Conclusions: Variability of the cardiovascular index W recovers more slowly after long-compared to short-duration spaceflight. Since high variability of W has also been shown to be predictive of OTT failure, a primary manifestation of OH, a logical conclusion is that recovery from OH also is slower after long-duration compared to short-duration spaceflights.

Feiveson, Alan H.↗
NASA NTRS1995

Gaussian statistics of the cosmic microwave background: Correlation of temperature extrema in the COBE DMR two-year sky maps

We use the two-point correlation function of the extrema points (peaks and valleys) in the Cosmic Background Explorer (COBE) Differential Microwave Radiometers (DMR) 2 year sky maps as a test for non-Gaussian temperature distribution in the cosmic microwave background anisotropy. A maximum-likelihood analysis compares the DMR data to n = 1 toy models whose random-phase spherical harmonic components a(sub lm) are drawn from either Gaussian, chi-square, or log-normal parent populations. The likelihood of the 53 GHz (A+B)/2 data is greatest for the exact Gaussian model. There is less than 10% chance that the non-Gaussian models tested describe the DMR data, limited primarily by type II errors in the statistical inference. The extrema correlation function is a stronger test for this class of non-Gaussian models than topological statistics such as the genus.

Kogut, A.↗